AperTO - Archivio Istituzionale Open Access dell'Università di Torino
Original Citation:
Cu-CHA-a model system for applied selective redox catalysis
Published version:
DOI:10.1039/c8cs00373d
Terms of use:
Open Access
(Article begins on next page)
Anyone can freely access the full text of works made available as "Open Access". Works made available under a Creative Commons license can be used according to the terms and conditions of said license. Use of all other works requires consent of the right holder (author or publisher) if not exempted from copyright protection by the applicable law.
Availability:
This is the author's manuscript
iris-AperTO
This is the author's final version of the contribution published as:
[Elisa Borfecchia, Pablo Beato, Stian Svelle, Unni Olsbye, Carlo Lamberti and Silvia
Bordiga, Cu-CHA – A Model system for Applied Selective Redox Catalysis, Chem.
Soc. Rev., 47, 2018, 8097-8133. DOI: 10.1039/C8CS00373D]
The publisher's version is available at:
https://pubs.rsc.org/en/content/articlelanding/2018/cs/c8cs00373d#!divAbstract
When citing, please refer to the published version.
Link to this full text:
http://hdl.handle.net/2318/1692622
Cu-CHA – A Model system for Applied Selective Redox Catalysis
Elisa Borfecchia,
a,†Pablo Beato,
*,aStian Svelle,
*,bUnni Olsbye,
bCarlo Lamberti
c,dand Silvia Bordiga,
*,b,ea Haldor Topsøe A/S, Haldor Topsøes Allé 1, 2800 Kongens Lyngby, Denmark. E-mail: PABB@topsoe.com
b Center for Materials Science and Nanotechnology (SMN), Department of Chemistry, University of Oslo, P.O. Box 1033 Blindern, N-0315 Oslo, Norway. E-mail: stian.svelle@kjemi.uio.no
c The Smart Materials Research Center, Southern Federal University, Sladkova 174/28, 344090 Rostov-on-Don, Russia d Department of Physics, INSTM Reference Center and CrisDi Interdepartmental Centre for Crystallography, University of Turin, Via P. Giuria 1, 10125 Turin Italy
e Department of Chemistry, INSTM Reference Center and NIS Interdepartmental Centre, University of Turin, Via Quarello 15, I-10135, Turin, Italy. E-mail: silvia.bordiga@unito.it
† Present address: SMN, Department of Chemistry, University of Oslo, P.O. Box 1033 Blindern, N-0315 Oslo, Norway.
We review the structural chemistry and reactivity of copper-exchanged molecular sieves with chabazite (CHA) topology, as industrially applied catalyst in ammonia mediated reduction of harmful nitrogen oxides (NH3-SCR) and as a general model system for red-ox active materials (also the recent
results in the direct conversion of methane to methanol are considered). Notwithstanding the apparent structural simplicity of the material, a crystalline zeolite with only one crystallographically independent T site, the Cu-SSZ-13 catalyst reveals a high degree of complexity that has been decrypted by state of the art characterization tools. From the reviewed data, the following important aspects in the understanding of the Cu-SSZ-13 catalyst clearly emerged: (i) the structural dynamics of the Cu-species require precise control of the environmental conditions during activation and characterization; (ii) the availability of a large library of well-defined catalysts with different Si/Al and Cu/Al compositional ratios is key in unravelling the red-ox property of the active Cu sites; (iii) a multi-technique approach is required, combining complementary techniques able to provide independent structural, electronic and vibrational information; (iv) synchrotron radiation based techniques (EXAFS, XANES, XES and time-resolved powder XRD) played a relevant role; (v) operando methodology (possibly supported by advanced chemometric approaches) is essential in obtaining structure-reactivity relations; (vi) the support of theoretical studies has been indispensable for the interpretation of the experimental output from characterization and for a critical assessment of mechanistic models.
The old literature, that classified Cu-exchanged zeolites in the category of single-site catalysts, has been partially disproved by the recent advanced studies where it has been shown that the active site in low temperature NH3-SCR catalyst is a mobile Cu-molecular entity that “lives in symbiosis” with
an inorganic solid framework. Only in the high temperature NH3-SCR regime, the mobile Cu-species
lose their ligands and find docking sites at the internal walls of the zeolite framework thus reflecting the idea of a single-site catalysts.
After a brief introduction, the review is divided into three main parts devoted to characterization (Section 2), reactivity (Section 3), and industrial applications (Section 4), followed by some concluding remarks and providing a perspective of the field.
1. Introduction
Natural chabazite is a tectosilicate mineral of the zeolite group, with a high aluminum content, that crystalize in the trigonal R-3m space group.1-3 Its synthetic counterpart can be obtained, by the use of
a sophisticated template, with Si/Al ratio higher than 10 and is known as SSZ-13.4 Finally, the same
topology can be obtained also as silicon aluminum phosphate, known as SAPO-34.5 The chabazite
topology is identified by layers of double six-membered rings (d6r) that are interconnected by units of four-membered rings (4r). The double six-membered-ring layers are stacked in an ABC sequence, leading to a framework with a regular array of barrel-shaped cages interconnected by eight-membered-ring windows (8r) (Fig. 1a). The chabazite framework is characterized by a high symmetry, as the asymmetric unit contains just one T site1-3, 6 and four non-equivalent oxygens, which
give origin to two families of Brønsted sites.7, 8 The first family include Brønsted sites associated with
O1, O2 and O4, that point inside the big cage and are easily accessible to reactants. The Brønsted sites
associated with O3 are instead less accessible, pointing inside the six-membered ring (6r) (Fig. 1b).
Fig. 1. Part (a): schematic representation of CHA topology, underlying the presence of 4r, 6r and 8r rings and the major dimension of the cages. Part (b): illustration of the four different oxygens available in the framework. Part (c): possibilities of isolated and paired Al sites, considered in the literature as most probable locations for Cu cations. Colour code: O red; Si grey; Al orange in case of zeolites (CHA or SSZ-13). A similar picture can be used to describe SAPO-34, where Al is replaced by Si and Si by alternating Al and P atoms. Previously unpublished figure.
A further element of complexity, in case of the CHA or the SSZ-13 zeolites, is represented by the Al loading, that implies a variability in the presence of one or two Al atoms in the zeolite rings. In particular, the different possibilities of isolated and paired Al sites are illustrated in Fig. 1c. Equivalent structures can be used to describe the SAPO-34 case,9 where the Al is replaced by Si and Si by
alternating Al and P atoms.
Apart from the interest in the acidic form of chabazites (H-CHA, H-SSZ-13 and H-SAPO-34),7, 8
mostly associated to the catalytic conversion of methanol into light olefins (MTO reaction),10, 11 and
in the H2-storage abilities,12, 13 chabazites shows some interest in catalytic properties engendered by
isomorphous substitution.14, 15 However, the major reason for the high research interest in inn the
system raised around 2008-2009, when the Cu-CHA redox catalyst was commercialized for NOx
reduction in lean-burn diesel engine aftertreatment. Moreover, the structural simplicity of the material attracted a great interest in Cu-CHA as a model system to resolve fundamental questions of structure-performance relationships in the general context of metal-exchanged zeolite catalysis.
In particular two relevant redox processes, where copper interconverts between Cu(I) and Cu(II) have been the subject of important research efforts in the last years: (i) the reduction of environmentally harmful NOx by NH3, i.e. selective catalytic reduction (SCR)16-27 and (ii) the direct
conversion of methane to methanol (MTM, sometime indicated in the literature as partial methane oxidation PMO).28-35
In order to shed light on the structural dynamics of the Cu-CHA system, it has been mandatory to apply advanced in situ/operando characterization approaches on a well-defined set of samples, by varying, in a controlled way, the measuring conditions.22, 36-40 Additional insights were provided by
the use of probe molecules, by performing ab-initio calculations on clusters and on periodic models, and by combining appropriate analytic tools, specifically developed to manage complex spectroscopic datasets. Fig. 2 illustrates the large variety of approaches employed in the literature to tackle the complexity of the problem, as will be exemplified by the studies discussed in this review.
Fig. 2. Scheme reporting the main experimental and theoretical tools used to shed light on the structure of the active Cu sites in CHA frameworks and their reactivity in the context of NH3 SCR and MTM reactions. Previously unpublished figure.
Starting from the bottom left, we can mention the use of: analytical tools such as ICP and gas-physisorption to define the sample composition and surface area. With respect to gas-gas-physisorption,41
the most used methods include N2- and Ar-physisorption, where the latter is often employed to
characterize the pore structure of CHA zeolites because of the incomplete packing of N2 in this zeolite
topology. Powder X-ray diffraction (PXRD) results in the phases identification needed to verify the purity of the crystalline phase. Moreover, if highly 2-resolved pattern (with a high signal-to-noise in the high-q-region) are available, PXRD can also be used to track the positions of the cations, thanks to a structure refinement procedure.6, 18, 42 Conversely, the short range order is mostly described by
extended X-ray absorption fine structure (EXAFS),37, 43-53 while spectroscopies involving electron
transitions (UV-Vis-NIR, luminescence, XANES and XES)27, 35, 45, 53-60 are able to discriminate
between Cu(I) and Cu(II) and to give insight on the local symmetry and coordinative state and the nature of coordinating ligands. Infrared spectroscopy61-63 provides a clear picture about the number
and the different families of Brønsted sites and, when used in combination with probe molecules, 64-67 can provide insight about the oxidation state and accessible coordinative environment of the Cu
species. Raman spectroscopy, mostly exploiting the resonance conditions,31, 68-74 allows the
identification of even very diluted surface species in controlled atmosphere. Theoretical calculations performed on both clusters and periodic models using DFT,24, 36, 53, 75-79 provide a relevant
contribution in the understanding of the local geometries and reactions paths. Molecular Dynamics (MD) enables a deeper mechanistic understanding under reaction conditions yielding adsorbate-mobilized Cu active-species (primary, low-temperature NH3-SCR, see Section 3.1.1), acting in a
pseudo-homogenous manner within the confined space of the zeolite cages.77, 79, 80
As Cu(II) is paramagnetic, EPR spectroscopy is widely used to characterize Cu(II) sites in exchanged zeolites.26, 81-88 or Cu(I) sites accessible to paramagnetic molecules such as NO.89 Solid
state NMR has been mostly used to give insight on 27Al local structure and in particular to identify
extra lattice aluminium sites.90, 91 In addition, 29Si was used to determine the distribution of Si(2Al),
NH3 dosage.92 Last but not least, operando facilities, starting from reaction tests, are more and more
frequently combined with spectroscopies, providing insight on the surface species formed and evolving under reaction conditions.22, 36-40
The present review is focused on Cu-SSZ-13 zeolites, as these are the materials on which the most abundant literature is available. The manuscript is divided into three main parts devoted to characterization (Section 2), reactivity (Section 3), and industrial applications (Section 4), that are finally followed by a section where conclusions and perspectives are drawn.
2. Structural complexity in the activated material
The coincidence of the structural simplicity of the CHA topology,1-3 together with the outstanding
performance of Cu-SSZ-13 in NH3-SCR, provoked a real boom of academic and industrial research
groups using Cu-SSZ-13 materials as model systems to address long-standing questions in metal-exchanged zeolite catalysis. Nonetheless, the vast literature accumulated in almost a decade of extensive research, reshaped the initial premise of a simple model material, progressively unveiling the complexity associated with metal speciation in the CHA framework.
Early reports18, 52, 93 described Cu-SSZ-13 as a ‘single-site’ catalyst, indicating an extra-framework
site in the plane of the 6r as the sole position for Cu ions in the dehydrated material. However, in the following years, it has become clear that the CHA topology offers multiple docking sites for Cu(II) and Cu(I) ions.6, 42, 53, 59, 75 Today it is well established that the chemical identity (oxidation state,
coordination geometry, nuclearity) and the framework location of Cu-species are determined by (i) physico-chemical environment (i.e. temperature, gas composition and gas flow rate) and (ii) sample composition (i.e. Si/Al ratio in the parent framework and Cu/Al ratio in the exchanged zeolite).53, 59, 75, 76, 94
This section addresses the current understanding of Cu-species formation and transformation in Cu-CHA materials as a function of two key factors: activation procedure (Section 2.1) and starting chemical composition (Section 2.2). Additional complexity is connected to the non-random distribution of Al in the zeolite framework and the playing room created by the possibility of on purpose modifying the Al distribution in the CHA framework (e.g. abundance of 6r hosting a pair of neighbouring Al atoms in −Al−Si−Al− or −Al−Si−Si−Al− coordination motifs)95 by tuning the
synthesis methods and parameters. Such a non-random distribution of framework Al has been known and described with some detail for ZSM-596 and other zeolites.97 However, the means by which to
selectively influence or even direct the Al distribution leaves something to be desired. Dedecek et al.96 varied the overall Al content, the sequence of mixing of reagents, the Si and Al source, as well
as the concentration of alkali cations and inorganic anions to prepare sets of ZSM-5 zeolites with different distributions of Al in the framework. It may be noted that ion exchange with [Co(II)(H2O)6]
followed by dehydration and analysis by UV spectroscopy provide a means to detect different Al arrangements.98 Dehydrated, bare Co(II) ions balanced by two close Al framework atoms give rise to
distinct d-d transitions, whereas various Co(II)-oxo species balanced by Al atoms further apart give rise to absorptions related to charge transfer transitions in the UV spectrum. Co(II) can also be used quantitatively as a titrant to determine the fraction of paired framework Al.99
A more systematic method to control the Al distribution has been reported for CHA. Di Iorio and Gounder95 demonstrated that SSZ-13 crystallized using the very phase selective trimethyladamantyl
ammonium cation from hydroxide resulted in only isolated Al atoms in the framework (Si/Al = 15-30). The addition of increasing concentrations of Na+ to the synthesis gel led to a linearly correlated
density of paired (separated by one or two Si units) framework Al. Moreover, it was demonstrated that at equimolar concentrations of SDA and Na+ it was possible to prepare a series of materials with
varying Al content for which the density of Al pairs was governed solely by a random distribution and Löwenstein’s rule.100 In a follow up contribution, the same team99 was able to show that two
neighboring Brønsted sites arising from paired Al were 10 times more active than isolated protons for the dehydration of methanol to DME.
However, the vast majority of contributions discussed here deal only with a limited number (one) of samples of zeolite supports, or do not characterize the Al distribution in detail (using the approaches based on the introduction of Co(II) mentioned above. Thus, it is in most cases not possible to discriminate between effects arising from the kinetics of the Cu exchange or by the density of paired Al. Further details on the synthesis influence on Cu-speciation can be found by the interested reader in the excellent chapter by Paolucci et al.,76 which exhaustively discusses this matter for
Cu-SSZ-13 and Cu-SAPO-34 catalysts.
In the following subsections 2.1 and 2.2, our focus here will be on the dehydrated (or ‘activated’) state of the material, which represents generally the starting point before catalysis. Herein we will demonstrate how in situ/operando multi-technique characterization guided by computational modelling has now enabled a renewed understanding of metal speciation and redox chemistry in these fascinating materials. The impact of the Cu-speciation in the activated material on the performance, as well as its reaction-induced modifications, will be then examined in Section 3 for two selected
processes, namely NH3-SCR and MTM conversion.
2.1. Activation-dependent Cu-speciation
Activation of zeolites, that are microporous materials, is a key step for both catalysis and sorption purposes, because the evacuation of water and other molecules adsorbed in the channels and cavities is necessary to make the active sites accessible to reactants. Incomplete activation can result in residual H2O or hydrocarbon molecules, which may block and/or modify the activity of a fraction of
the sites.63, 101 The same holds when activation precedes a volumetric, calorimetric or spectroscopic
investigation with a specific probe molecule; the simultaneous presence of water may prevent adsorption of the probe and/or may result in misleading results because of competition for the adsorption at the same site.
Since the discovery in the nineties that Cu-ZSM-5 zeolites are active in the direct decomposition of NOx into N2 and O2,102-108 the activation of Cu-exchanged zeolites has been followed by several
spectroscopic studies. Besides isolated cases where copper was inserted in the zeolite as Cu(I) species via gas-phase46, 48-50, 109-113 or solid phase114-116 exchanges with CuCl, in the vast majority of the
literature the cation exchange was performed via aqueous phase ion-exchange using different cupric salts.
The nature of copper in as-prepared ion-exchanged zeolites from aqueous preparation conditions, is generally agreed to be hydrated Cu(II) species. It was also immediately recognized that hydrated Cu(II) species in copper-exchanged zeolites undergo a progressive reduction to Cu(I) upon activation under vacuum or inert atmosphere (He or N2) at increasing temperatures.82-84, 117-122 The phenomenon
has been described as “self-reduction” or “auto-reduction”, owing to the fact that even though several different reduction processes were proposed, no direct proof or identification of the species that has to become oxidized in the process was found.
On the other hand, already in the middle of the nineties it became evident that an activation performed in presence of O2 resulted in samples characterized by very different copper species,123, 124
as compared to an inert or vacuum activation. This evidence was confirmed in successive works.53, 59, 65, 72, 73, 125 This was actually an important finding, since in a catalytic redox process both states, the
reduced and oxidized metal state, should be addressed. In this regard, the collaboration between Haldor Topsøe and different academic partners provided a systematic investigation of three batches of Cu-SSZ-13 samples prepared in the same way with comparable chemical composition (13.1 ≤ Si/Al ≤ 15.5 and 0.44 ≤ Cu/Al ≤ 0.48) subjected to both inert- and O2-activation using an impressive
number of complementary characterization techniques: IR,53, 65 IR of adsorbed probe molecules,58, 65
UV-Vis,65 EPR,65, 86 XANES,53, 58, 59 EXAFS,53, 59 XES,53, 58 synchrotron radiation PXRD6 and
simultaneous synchrotron radiation PXRD/XANES.42 The picture was then also supported by
quantum mechanical modelling using DFT.6, 53, 59 Hereafter we will mainly summarize the results of
those combined studies as they represent a set of self-consistent data sets coming from independent spectroscopic and scattering techniques, providing a complete structural, electronic and vibrational
picture on the evolution of the population of Cu sites along the two different activation procedures.6, 42, 53, 58, 59, 65, 86
2.1.1. IR spectroscopy: revealing the presence of Cu(I) and Cu(II) sites and of [CuOH]+ species.
Starting form IR spectroscopy, Fig. 3a reports the FTIR spectra of dehydrated Cu-SSZ-13 zeolite collected after O2-activation and vacuum-activation, red and black curves, respectively. The main
features in the ν(O–H) stretching region, left part, are the bands due to external silanols (3737 cm1)126-128 and to internal Brønsted sites (3611 cm1, 3584 cm1),7, 65 that are similarly present after
both activations. In the low wavenumber region (right part of Fig. 3a), the main feature in the 840-770 cm1 region, which is due to the total symmetric stretching of the [SiO4] units,68 dominates both
spectra. However, the spectrum of the O2-activated sample exhibits two minor, but distinct, features
at 3656 and 905 cm1 which are not observed for the vacuum activated sample. Giordanino et al.65
were the first to assign those bands to the stretching and bending fingerprints of [CuOH]+ species
stabilized in the SSZ-13 matrix,53, 65 assignment successively confirmed by authoritative groups
active in the field,35, 75, 129 vide infra Fig. 5a for a pictorial representation of this Cu(II) species hosted
in the large cage of the CHA framework. The absence of both bands in the vacuum-activated sample allowed Borfecchia et al. to conclude that the stabilization of the OH extra-ligand on Cu(I) does not occur.53
Fig. 3. Low temperature (~100 K) IR spectra of O2-activated (red) and vacuum-activated (black) Cu-SSZ-13 zeolite (Si/Al = 13.1 and Cu/Al = 0.444). Part (a): effect of activation at 400 °C in the ν(OH) and δ(OH) stretching regions, left and right parts, respectively. For comparison, also the spectrum of the vacuum-activated H-SSZ-13 material (gray) is reported. The O2-activated zeolites presents the fingerprint bands of [CuOH]+ species at 3655 and 905 cm1. Part (b): IR spectra of CO adsorption at low, intermediate and high PCO, bottom, middle and top curves respectively, approximately corresponding to the saturation of mono- (M band), di- (D doublet) and tri-carbonyl (T triplet) complexes. At the highest PCO, also the bands of CO adsorbed on Brønsted (B), silanols (S) and the band of liquid-like CO are present after both activation. Part c: IR spectra of NO adsorption at low and high PNO, bottom and top curves respectively, approximately corresponding to the saturation of mono- and di-nitrosyl complexes on Cu(I) sites. In all parts some spectra have been vertically shifted for clarity. Unpublished figure reporting spectra previously published in Ref.53 for part (a) and in Ref.65 for parts (b) and (c).
The comparison with the spectrum collected on the non-exchanged parent H-SSZ-13 material (gray curves in Fig. 3a) highlighted that the intensities of the 3611 and 3584 cm1 bands related to
Brønsted sites are only slightly lower in the Cu exchanged sample (red or black spectra). This spectroscopic observation was particularly relevant because it could explain why a sample which has been theoretically almost fully ion-exchanged (Cu/Al ratio of 0.444) can still preserve plenty of free Brønsted acid sites. The formation of [CuOH]+ species was explained according to two possible
mechanisms.53
In case two framework Al atoms are in a close proximity (2Al Z2 sites), the stabilization of divalent
[Cu(II)(H2O)n]2+ complexes is favored and their progressive dehydration leads to H2O dissociation
adjacent framework Al atoms. The FTIR study of Borfeccia et al.53 proved that, at 250 °C, copper
sites are fully dehydrated and for higher temperatures [Cu(II)OH]+ species can only be stabilized in
an oxidative atmosphere only, otherwise they undergo “self-reduction” as a consequence of OH
extra-ligand loss. Alternatively, dehydration of [Cu(II)(H2O)n]2+ complexes could lead to bare Cu(II)
cations, Eq. (1b). Conversely, in those sites characterized by only one Al in the proximity (1Al Z sites), the hydrated state upon aqueous ion exchange is a monovalent [Cu(II)(H2O)n(OH)]+ complex,
that does not require any water dissociation to be transformed into [Cu(II)OH]+ upon dehydration,
see Eq. (2), and the concentration of Brønsted sites in the dehydrated material is the one predicted from a total exchange level of [Cu(II)OH]+/Al3+ = 1. In all these cases, the loss of the OH
extra-ligand results in the reduction of the Cu(II) center to Cu(I).53
2𝐴𝑙 Z : [Cu(II)(H O) ] → [Cu(II)(H O)] + (𝑛 − 1)(H O)↑ → [Cu(II)OH] + H →
→ Cu(I) + (OH)↑+ H (1)
2𝐴𝑙 Z : [Cu(II)(H O) ] → Cu(II) + 𝑛(H O)↑ (1a)
1𝐴𝑙 Z: [Cu(II)(H O) (OH)] → [Cu(II)OH] + 𝑛(H O)↑ → Cu(I) + (OH)↑ (2)
The reversibility of the OH extra-ligand loss has been confirmed by XAS and FTIR, demonstrating
that Cu(I) sites rapidly undergo re-oxidation with consequent restoration of [Cu(II)OH]+ species if
they are exposed to a gas mixture of O2/H2O.53
IR spectroscopy of adsorbed probe molecules is a powerful technique in the characterization of surface sites of porous, high surface area, materials.61-64, 130, 131 The choice of probe molecule is a
crucial point in this approach; different probe molecules may be able to reveal different aspects of the investigated surface, and often the combined use of markedly different probes is the key to reach a comprehensive understanding of the surface.64 In this regard, Giordanino et al. reported an extensive
characterization of Cu-SSZ-13 activated under oxidative and reductive (vacuum) conditions and tested by N2, CO and NO probes.65
CO is the ideal molecule to probe Cu(I) sites because of both the high extinction coefficient and the strong adsorption energy.111, 132 The high coordinative unsaturation of Cu(I) cations hosted in
vacuum-activated zeolites was already demonstrated in the 1990’s by the observation that multiple carbonyl complexes form, i.e. Cu(I)(CO)n (n =1,2,3), depending on CO equilibrium pressure (PCO)
and temperature.45, 46, 49, 50, 84, 110, 133, 134 At low P
CO (bottom spectra in Fig. 3b) the typical band (M) at
2155 cm1 due to Cu(I)(CO) monocarbonyl complexes is observed. By increasing PCO the doublet at
2178 and 2148 cm1, due to the symmetric and asymmetric stretching modes of the Cu(I)(CO)2
complexes,45, 46, 49, 50, 84, 109, 110, 133, 134 (D-bands, in the middle spectra of Fig. 3b) start to appear at the
expense of the monocarbonyls (M-band). M- and D-bands appear independently on the activation treatment but are significantly more intense (by a factor 4) in case of inert-activated sample (black curves), confirming the higher concentration of Cu(I) ions relative to the O2 activated one (red
curves). In addition, a more heterogeneous distribution of the Cu(I) sites after oxidative pretreatment compared to inert-activation is recognized by comparing the FWHM of the M bands, which are 14 and 9 cm1, respectively.
Higher PCO leads to the formation Cu(I)(CO)3 tri-carbonyl complexes (triplet T at 2134, 2169 and
2194 cm1, in the top spectra in Fig. 3b), that are however less clearly visible because of an increased
complexity in the IR spectra. Indeed, at such PCO values, CO starts to be coordinated also on sites
characterized by lower adsorption enthalpies such as Brønsted and silanols, bands at 2173 cm1 (B)
and 2160 cm-1 (S), respectively126, 135. The band at 2138 cm1 is due to liquid-like CO physisorbed in
the zeolite channels.135-137 Giordanino et al.65 noticed that the intensity of the Brønsted acid sites
related bands is not enhanced in the vacuum activated sample compared to the O2 activation. This
indicates that for a Cu-CHA sample with Si/Al~15 and Cu/Al~0.5 the self-reduction of Cu(II) into Cu(I) during the vacuum activation process does not generate new H+ sites to compensate the loss of
one positive charge for all copper atoms. This observation rules out the presence of a significant fraction of Cu(II) able to balance the negative framework charge induced by two adjacent Al atoms
and implies that most of the cupric sites are inserted in form of monovalent complexes such as [Cu(II)OH]+.65
Due to the weaker interaction between Cu(II) and CO the corresponding Cu(II)(CO) complexes are very unstable and difficult to detect even at low temperature,131, 138 particularly when they
co-exist with even just a minority of Cu(I) sites. This is the reason why the IR band of Cu(II)(CO) complexes, expected around 2194 cm1, is not appreciable, even in the O2-activated sample, where
Cu(II) surface species are the most abundant ones. Cupric species can be monitored by NO as a probe, with the advantage that NO is able to form complexes with both Cu(I) and Cu(II) sites.45, 61, 64, 67, 112, 139, 140 At low P
NO, vacuum activated Cu-SSZ-13 (black curve in the bottom of Fig. 3c) shows a band
centered at 1810 cm1 due to Cu(I)(NO) mononitrosyl complexes;45, 61, 64, 109, 112, 139 At high P
NO (top
black curve), the 1810 cm1 band evolves in two main components at 1826 and 1728 cm1 due to the
symmetric and asymmetric stretching of Cu(I)(NO)2 dinitrosyl complexes, respectively.45, 61, 64, 112, 139, 140 The absence of any significant band in the 1950-1870 cm1 range, where Cu(II)(NO) complexes
are expected, implies that almost all cupric ions in accessible positions have been reduced to cuprous ions during the thermal activation in vacuo. The situation is different for the O2-activated sample (red
curves in Fig. 3c), where bands due to Cu(I)(NO) and Cu(I)(NO)2 complexes are present as a
minority, and the spectra are dominated, at both low and high PNO, by a complex and structured
absorption in the 1970-1850 cm1 range due to Cu(II)(NO) adducts. The complexity of the absorption
features of Cu(II)(NO) adducts indicates the presence of different Cu(II) cationic sites, that are not present in the reduced cuprous sites.65 The heterogeneity of Cu(II) sites was also confirmed by the
combined H2-TPR/IR study of Kwak et al.141 and by the EPR study of Godiksen et al.,86 who found
two EPR active sites in the 6r. In the same work the authors also addressed the long-standing question of the EPR silent monomeric Cu(II) species in Cu-zeolites82-84 in terms of [Cu(II)OH]+ species
coordinated to two framework oxygen atoms bonded to an isolated Al atom, namely 1Al Z[Cu(II)OH] according to the nomenclature introduced above.86 CO and NO molecules are coordinated too
strongly to Cu(I) to be suitable probes to discriminate among small cuprous site inhomogeneities; weakly interacting molecules such as N2 are required for this porpoise. Giordanino et al.65 followed
the evolution of IR spectra of when dosing N2 at around 100 K on vacuum activated Cu-SSZ-13,
observing two components at 2293 and 2300 cm−1. The two bands were assigned to N
2 molecules
adsorbed on Cu(I) sites in the 8r and 6r respectively,59, 65 the former being a stronger adsorption site,
because the 2300 cm−1 band is the only one observed at low P
N2, and because it is characterized by a
larger red-shift of the (NO) stretching mode. The observation that the relative intensity of the IR components of the two Cu(I)N2 complexes changes in Cu-SSZ-13 samples characterized by
different Cu/Al and Si/Al ratios clearly indicated a dependence of Cu(I) site distribution on the catalyst composition and will be commented in Section 2.2, vide infra Fig. 13b.
2.1.2. UV-Vis and Raman spectroscopies: insights on Cu(II) active-oxygen species. Very peculiar are the d-d transitions of Cu(II) species in O2-activated Cu-SSZ-13, see the UV-Vis-NIR spectrum in
Fig. 4a (blue curve). Giordanino et al.65 highlighted the presence of a very intense and very well
defined quadruplet with maxima at 19700, 16500, 13600 and 11000 cm1, that is responsible for the
deep blue color of the O2-activated sample, see inset in Fig. 4a. Still under debate is the precise
assignment of this quadruplet either to different Cu(II) species/sites, reflecting the Cu(II) heterogeneity discussed above, or to a single site subjected to a very specific ligand field.86 Support
from advanced DFT calculations is needed here. The sample is also characterized by a complex charge-transfer band extending down to 25000 cm1. Oord et al.35 reported operando UV-Vis-NIR
spectra following the sample activation, both in O2- and He-flow conditions, see Fig. 4b and c,
respectively. Upon activation the typical d-d component of hydrated Cu(II) species around 12000 cm1 (green spectra in Fig. 4a,b,c) progressively evolves into the structured quadruplet discussed
above, which intensity however does not show a monotonic increase with activation temperature as it reaches a maximum at 320 °C and then partially loses intensity (see arrows in Fig. 4b,c). This observation is in line with the in situ FTIR study by Pappas et al.,33 who observed that the finger print
band of cupric [Cu(II)OH]+ species was depleted by increasing the O
250 °C (Fig. 4d). The activation experiments by Oord et al. were successively followed by methane addition during temperature ramp from 60 to 200 °C (vide infra section 3.2.4, in particular Fig. 22e); the authors concluded their operando UV-Vis-NIR study observing that an oxidizing agent is needed to create the active site for methane to methanol activation in Cu-SSZ-13 as no methanol production was measured after He-activation.35
Fig. 4. Part (a): Comparison between in situ UV-Vis-NIR spectra of hydrated (green curve) and O2-activated (blue curve) Cu-SSZ-13 zeolite (Si/Al = 13.1 and Cu/Al = 0.444). Adapted by permission of the Royal Chemical Society (copyright 2013) from ref. 65 Part (b): operando UV-vis spectra of the Cu-1.21-Na-Cu-SSZ-13 zeolite (Si/Al = 20 and 1.21 Cu wt %) during O2-activation from RT (green spectrum) to 450 °C (red spectrum). The arrows indicate the time evolution, the green spectrum is the starting point of the experiment, the d-d quadruplet reaches maximum intensity at 320 °C, blue spectrum. Part (c): as part (b) during He-activation. Parts (b,c) adapted with permission of the Royal Chemical Society (copyright 2018) from ref.35. Part (d): FTIR spectroscopy in the ν(OH) stretching region as monitored in situ along O2-activation of Cu-SSZ-13 (Si/Al = 14.8 and Cu/Al = 0.53) catalyst in the 250 – 300 °C range. The red arrow highlights the progressive decrease in the intensity of the band at ca. 3650 cm−1 fingerprint band of cupric [CuOH]+ species upon increasing the activation temperature from 250 °C. Part (e): Raman spectra (ex = 488 nm) of Cu-SSZ-13 (Si/Al = 14.8 and Cu/Al = 0.53) catalyst in its hydrated (green line) and O2-activated (blue line) forms. The spectral contributions from the observed Cu(II)xOy moieties are highlighted: [Cu(trans-μ-1,2-O2)Cu]2+ (pink shading); [Cu-(μ-O)-Cu]2+ (orange shading); and [Cu(II)O2•]+ (light-blue shading). Parts (d,e): adapted by permission of the American Chemical Society (copyright 2017) from ref.33
Raman spectroscopy is particularly suited to detect multinuclear Cu(II)xOy, it has been widely used
to characterize active copper sites in biological systems57, 142 and in O
2-activated zeolites.31, 33, 72-74, 143 Fig. 4e reports the evolution of the Raman spectrum of Cu-SSZ-13 catalyst before (green line) and
after (blue line) O2-activation.33
Several new vibrational features, related to oxygen-activated Cu(II) species appear after O2
-activation: [Cu(II)O2•]+ end-on superoxo species (strong and complex light-blue shaded components
in the 1000 - 1200 cm−1 interval, with two maxima at 1100 and 1155 cm−1)33, 72 in equilibrium with
the corresponding side-on species, pictorially represented in Fig. 5b;33 [Cu(trans-μ-1,2-O
2)Cu]2+
complex (orange shaded components at 510, 580, and 830 cm−1), see Fig. 5c for a pictorial
Fig. 5. Pictorial representation of some proposed local structures of the copper species formed in the large cage of the CHA frameworks under oxygen treatment at high temperature. (a): [CuOH]+ species. (b): [Cu(II)O2•]+ side on superoxo species. (c): [Cu(trans-μ-1,2-O2)Cu]2+. Colour code: Si grey, O, red, Al gold, Cu green, H, white. By courtesy of K. P. Lillerud (Department of Chemistry, University of Oslo), previously unpublished Figure.
2.1.3. DFT supported XAS and XES investigation: understanding the structure of the Cu(II) and Cu(I) species inn activated samples. Borfecchia et al.53 followed the activation in temperature
up to 400 °C in both O2/He and pure He flows with XANES, EXAFS and valence-to-core XES
spectroscopies. The EXAFS-optimized experimental setup and rather high copper content in the sample allowed a very good data quality paving the way to a detailed quantitative analysis. XANES, EXAFS and valence-to-core XES spectra, together with the optimized DFT models and corresponding XANES and XES simulations are reported in Fig. 6 and Fig. 7 for O2- and He-
activated samples, respectively. Borfecchia et al.,53 tested several DFT models for Cu(II) and Cu(I)
sites in the 8r or in the 6r, obtained inserting either one or two Al atoms in the T positions of the rings. Structures reported in Fig. 6c and Fig. 7c,d are only those that are compatible with the experimental results.
Fig. 6. Part (a): XANES spectra following the activation from room temperature (blue curve, hydrated material) to 400 °C (red curve, activated material) of Cu-SSZ-13 (Cu/Al=0.444, Si/Al=13.1) in 50% O2/He flow. The inset shows a magnification of the 1s3d transition, typical of Cu(II) species. Part (b) as part (a) for the k2-weighted FT of the EXAFS spectra. Part (c): DFT model of the dominant Cu-site in the O2-activated material. Part (d): best EXAFS fit and related main individual components obtained using the model reported in part (c). Part (e): experimental HERFD XANES spectrum (black curve) and computed XANES spectra (colored curves) for the different optimized possible sites. Part (f):
as part (e) for the valence to core XES spectra. Both HERFD XANES and XES simulations support the EXAFS results. Adapted by permission of the Royal Society of Chemistry (copyright 2015) from ref.53
Upon activation in O2/He flux, Cu(II) centers undergo progressive dehydration, while interacting
more closely with the framework, maintaining the +2 oxidation state. Features typical for Cu(II) in low-symmetry environment were observed in XANES (Fig. 6a), while EXAFS witnesses the marked decrease of the first shell intensity due to the loss of the coordinated water molecules (Fig. 6b). Comparable evolution of the XANES spectra upon similar activation in oxidative environment was observed also by Kwak et al.38 Conversely, as already observed by IR spectroscopy studies (see
Section 2.1.1), upon activation in vacuum or in inert atmosphere the Cu oxidation state changes to +1, as evidenced for the Cu(II)-SSZ-13 system by the disappearance of 1s3d transition and by the additional redshift of the edge, see Fig. 7a. Most interestingly, the EXAFS data reveal that the coordination of Cu upon He-activation was further decreased compared to the activation in O2.
Coupled with the observation that the reduction in He flow appears only at high temperature (T > 250 °C), while at lower T the evolution of the spectra is identical to the O2-activation case, it indicates
that a charged extra-ligand is still coordinated to Cu even at high temperature in case of O2-activation.
This evidence supports the hypothesis of the presence of an OH ligand in the first coordination shell
of Cu(II) as advanced in the IR study of Giordanino et al.65 to assign the ν(OH) stretching mode at
3657 cm−1 and discussed in the previous section.
Fig. 7. Part (a): XANES spectra following the activation from room temperature (blue curve, hydrated material) to 400 °C (red curve, activated material) of Cu-SSZ-13 (Cu/Al=0.444, Si/Al=13.1) in inert He flow. The inset shows a magnification of the 1s3d transition, typical of Cu(II) species and disappearing at high temperature. Part (b) as part (a) for the k2-weighted FT of the EXAFS spectra. Parts (c,d): DFT model of the dominant Cu-sites in the He-activated material. Best EXAFS fit and related main individual components obtained using the model reported in part (c). Part (d): experimental HERFD XANES spectrum (black curve) and computed XANES spectra (colored curves) for the different optimized possible sites. Part (f): as part (e) for the valence to core XES spectra. Both HERFD XANES and XES simulations support the EXAFS results. Adapted by permission of the Royal Society of Chemistry (copyright 2015) from ref.53
Fig. 8. Schematic representation of temperature/condition dependent Cu-speciation in Cu-SSZ-13 (Cu/Al=0.444 and Si/Al=13.1). O2-activation resulted in a virtually 100 % Cu(II) sample, while the presence of a minor fraction (around 10%) of Cu(II) was detectable in the He-activated sample using HERFD XANES spectroscopy. The larger geometries depicted in the right side of the scheme are the dominant structural components identified by XAS, XES and FTIR in the Cu-SSZ-13 sample with this composition. As indicated in the scheme, based on the UV-Vis and Raman results discussed in Section 2.1.2 (Fig. 4), high-temperature thermal treatment in the presence of O2 could also result in the formation of Cu(II)/active-oxygen species, see Fig. 5. Adapted by permission of the Royal Society of Chemistry (copyright 2015) from ref.53
Borfecchia et al.53 have further tested this hypothesis by performing a set of DFT simulations of
Cu ions in different locations of the framework and using the resulting structures as input for EXAFS fits and for the simulations of the high energy resolution fluorescence detected (HRFD) XANES and XES spectra, see parts (e) and (f) of Fig. 6 and Fig. 7. For the O2-activated material, the best overall
agreement with the experimental data was obtained for the models of Cu(II) in the 8r, in form of a Z[Cu(II)OH] complex, see Fig. 6c confirming the first assignment of the ν(OH) stretching mode at 3657 cm−1,65 while in case of He-activation it was a bare Cu(I) cation hosted mainly in the 8r, with a
minority occupancy of the 6r site, as summarized in the scheme reported in Fig. 8. This study was then extended in a successive work of the same group59 to a set of six Cu-SSZ-13 samples
characterized by different Si/Al And Cu/Al ratios, that will be discussed hereafter in Section 2.2.3, see Fig. 12.
2.1.4. Combined XANES/PXRD: highlighting the correlation between Cu(II) self-reduction and cation migration in the CHA framework. As discussed in the previous section, the in situ XAS/XES study by Borfecchia et al.53 shed light on the evolution of the oxidation state and local environment
of copper species along the two different activation procedures. Moreover, the support of DFT, allowed the use of the EXAFS spectra as an indirect tool to determine whether Cu species are located in the 8r or 6r, see Fig. 12. A direct determination of the copper location in the unit cell requires however diffraction techniques. In this regard, Andersen et al.6 provided a detailed structural
description of an O2-activated Cu-SSZ-13 (Si/Al=15.5, Cu/Al=0.45) analyzing with iterative Rietveld
analysis and maximum entropy method high-resolution synchrotron PXRD, finding that Cu cations occupied two crystallographic independent sites located in the 8r and in the 6r. The limitation of the first study by Andersen et al.6 is however that information on the oxidation state of copper ions cannot
be extracted from diffraction data.
To overcome this limitation, Andersen et al.42 performed a new experiment on the
Swiss-Norwegian beamline of the ESRF synchrotron that is equipped with two independent monochromators, allowing a rapid (about 30 s) plug and play switch between X-ray absorption and X-ray diffraction setups.144-147 Thus both XAFS spectra and PXRD patterns (with a 2D detector) can
Using the two Cu sites determined in the previous experiment,6 and constraining the sum of the
refined Cu species to be equal to 1.08 atoms per unit cell (as determined by ICP elemental analysis) Andersen et al.42 found that along the O
2-activation, there is a migration of Cu species from the 8r
into the 6r site (black and red data, respectively in Fig. 9a) and that this phenomenon is accompanied by a partial reduction of Cu(II) into Cu(I) (violet and orange data in Fig. 9b). This reduction phenomenon occurs once virtually all water molecules have left the zeolite framework, as confirmed by the in dependent PXRD and XANES analyses, blue data in parts (a) and (b) of Fig. 9, respectively. The reduction of a fraction of Cu(II) into Cu(I), determined by linear combination analysis of the XANES data, is essential to explain how, at the end of the treatment, copper species can occupy the 6r site with an occupancy almost twice as large as the maximum allowed for Cu(II) species on the basis of the Si/Al ratio of the investigated zeolite (15.5). The same experiment, repeated in He-atmosphere (Fig. 9c,d), resulted in a more pronounced migration of copper species from the 8r into the 6r site and in a total reduction of Cu(II) into Cu(I), regardless of which site is occupied.
Fig. 9. Part (a): Evolution of the Cu occupancies in the 8r (black data) and 6r (red data) sites during O2-activation (10% O2 in He) of Cu-CHA (Si/Al = 15.1; Cu/Al = 0.48) from the Rietveld refinement of the time-resolved synchrotron PXRD data. In all refinements, the sum of the occupancies of the two sites has been constrained to the value determined by chemical analysis (dashed vertical gray line). Also reported is the number of water molecules per unit cell optimized in the refinements (blue data). The horizontal dashed orange line represents the theoretical maximum amount of Cu(II) species that can be hosted in 6r in a CHA framework with Si/Al = 15.5. Part (b): fraction of hydrated Cu(II) species (blue data) of Cu(II) species interacting with the zeolite framework (violet data) and of Cu(I) species interacting with the zeolite framework (orange data) as determined by the linear combination analysis of the XANES data collected almost simultaneously with the PXRD data. Parts (c) and (d): as parts (a) and (b) for the He-activation of the same sample.
Comparing the overall results of the IR experiments of CO and NO adsorption, (Section 2.1.1, Fig. 3) with the combined XANES/EXAFS/XES experiments (Section 2.1.3, Fig. 6, Fig. 7 and Fig. 8) and with the parallel PXRD/XANES study reviewed here above (Fig. 9), it emerges that all experiments are in semi-qualitative agreement and that some (minor) disagreement exists on the quantification of the fraction of Cu(I) and Cu(II) species obtained at the end of both O2- and
inert-activations. To better understand such discrepancies the next section extends the analysis of the complexity of the problem to two important variables that are the Si/Al and the Cu/Al ratios.
2.2. Composition-dependent Cu-speciation
2.2.1. Multiple cationic positions in Cu-CHA: insight and compositional trends from TPR. As anticipated in the beginning of Section 2, the initial structural investigations on Cu-CHA zeolites3, 93
claimed that Cu is located in a single cationic site, coordinated to three framework oxygen atoms (Ofw) just outside the 6r plane. In particular, the study by Fickel et al.93 was conducted on a
Cu-SSZ-13 catalyst with Si/Al = 12 and Cu/Al = 0.35 by Rietveld refinement of temperature-dependent synchrotron XRD data. A subsequent report by Korhonen et al.52 supported the previously proposed
UV-Vis and XAS data. As a consequence, the isolated Cu(II) species in 6r identified by XRD were initially proposed as the active sites for NH3-SCR.
This simple and elegant structural picture was however challenged by the TPR results reported by Peden and co-workers.141 The authors performed H
2-TPR measurements on a series of
Cu-exchanged SSZ-13 zeolites, with Si/Al2 = 12 and exchange levels from 20% to 100%. In particular,
H2-TPR experiments performed in 2% H2/Ar in the RT–600°C temperature range on O2-calcinated
catalysts evidenced only a single H2 consumption peak at 340 °C for the 20% exchanged zeolite.
Increasing the Cu-exchange level, an additional peak developed at 230 °C, the intensity of which increased proportionally to the Cu-loading, and was maximized at 100% Cu-exchange. As for the high-temperature peak at 340 °C, its intensity was observed to be stabilized at 40% exchange, remaining unchanged at higher Cu-loadings. These findings provided a direct proof of two different Cu(II) species/sites in SSZ-13, with Cu-loading-dependent distribution and markedly different redox barriers. In the same work, the authors also explored the effect of H2O presence during TPR, working
in 2%H2/Ar+1% H2O. Under these conditions, the two reduction peaks observed during ‘dry’ TPR
would progressively shift towards each other as Cu-loading increased, coalescing finally to a single peak at 210 °C for the 100%-exchanged sample.
These findings, together with complementary evidence from FTIR spectroscopy of adsorbed CO and NO (see above the discussion of Fig. 3b,c), were interpreted with two types of Cu ions in Cu-SSZ-13: (i) Cu in highly coordinated and stable sites in the 6r, proposed to be primarily occupied at low exchange level and associated with the high-temperature TPR peak; (ii) Cu in the large cages of the CHA framework responsible for the low-temperature H2 consumption peak and favored at higher
exchange levels. Interaction with H2O, even in very small amounts, was suggested to cause migration
of the Cu ions, driving Cu out from the 6r site towards the large CHA cages, where it could be more easily reduced.
A subsequent study by the same research group148 comprehensively addressed the impact of both
Cu/Al and Si/Al ratios on the redox behaviour of Cu-SSZ-13 catalysts as probed by H2-TPR. In this
work, Gao et al. collected H2-TPR data for a large set of samples, with Si/Al ratios of 6, 12 and 35
and several different Cu/Al ratios in the 0.06−0.44 range. Fig. 10 reports an overview of the results. The materials were analysed starting from their hydrated state (‘hydrated’ labels in Fig. 10) and after being pre-dehydrated in dry 5% O2/He at 550 °C (‘dehydrated’ labels in Fig. 10).
The high-temperature H2 consumption peak is favoured at low values of both Si/Al and Cu/Al
ratios. It becomes almost undetectable for Si/Al = 35 catalysts, where the low-temperature peak dominates the TPR profile, irrespectively of the Cu-loading. For the hydrated catalysts a general enhancement of the low-temperature peak is observed. Notably, at high Cu/Al ratio, dehydrated samples exhibited a significant shift of the first reduction peak to lower temperature relative to the hydrated ones.
The authors also found that the total H2 consumption for the two highest Cu loadings diminished
for dehydrated samples with respect to the hydrated ones. This observation was interpreted with self-reduction of some Cu(II) already during dehydration step prior to TPR, even if the treatment was carried out in a 5% O2 /He flow. Evidence for self-reduction of about 40% of the total Cu content
during thermal treatment in diluted O2 flow (10% O2/He) at temperature > 450 °C was also reported
by Andersen et al.,42 see above Section 2.1. The integrated area of the TPR peaks for the dehydrated
samples with Si/Al = 35 where connected with the highest self-reduction levels, showing a peculiar decrease of self-reduced Cu at increasing Cu-loading.
Fig. 10. H2-TPR results for Cu-SSZ-13 samples with different Cu loadings for (a) Si/Al = 6, (b) Si/Al = 12 and (c) Si/Al = 35. Top panels report TPR for dehydrated samples while bottom panels show TPR for fully hydrated samples with the same composition. Adapted with permission from ref.148. Copyright 2015 Elsevier Inc.
It clearly appears that the redox-properties of the Cu-ions in Cu-SSZ-13 can be systematically tuned by modifying the compositional parameters of the material. The TPR results reviewed above point to two major Cu(II) species (or, possibly, classes of Cu(II) species). These include a redox-resistant componentfavoured at low Si/Al and Cu/Al values and a redox-active one, becoming more abundant at high Si/Al and Cu/Al ratios. Based on the TPR trends with composition, the redox-resistant component is consistent with bare Cu(II) species charge-balanced by two proximal Al atoms at T-sites in a 6r, Z2Cu(II), statistically more abundant in Al-rich frameworks, see above Eq. (1a) and
related discussion. Highly stable and firmly coordinated to the framework, this site well matches the initial proposals formulated from XRD analysis. The assignment of the redox-active component, based on TPR, appears somehow more uncertain. Its composition-dependent H2-TPR response
strongly supports a Cu(II)species hosted at a 1Al sites, where the charge balance is closed by extra-framework ligands. The Z[Cu(II)OH]+ complex extensively discussed in Section 2.1 represents a very
plausible candidate. Nonetheless, the differences among hydrated and dehydrated conditions evidenced in Fig. 10 indicate a role of high-temperature treatment in O2. According to Gao et al.148 a
reduction onset as low as 100 °C for materials pre-activated in O2, suggests the presence of dimeric
oxo-bridged Cu-species, or possibly other O2-derived superoxo or peroxo moieties, undergoing facile
reduction even at very low temperature. The formation of such types of species, already introduced in Section 2.1.2, will be further discussed in sub-section 2.2.4.
2.2.2. A compositional phase diagram for Cu-SSZ-13. The TPR results described above where paralleled by an impressive amount of characterization results on Cu-CHA materials at reference composition (typically Si/Al ~ 12–15, Cu/Al ~ 0.5, giving optimal performance in NH3-SCR) under
different activation conditions (see Section 2.1). In 2016, a milestone work from Schneider and co-workers75 synergized theory and experiment to rationalize in a consistent picture the abundant
Fig. 11. (a) Theoretical compositional phase diagram for Cu-sites in O2-activated Cu-SSZ-13. Color scale on the right indicates the predicted fraction of redox-active 1Al Z[Cu(II)OH] species as a function of Cu/Al and Si/Al ratios. The white line demarcates the transition from 2Al Z2Cu(II)-only region to the mixed 2Al Z2Cu(II) / 1Al Z[Cu(II)OH] region. Structural models of the two Cu-species are also reported (atom colour code: Cu, green; O, red; Si, grey, Al, orange; H, white). Full white dots indicates compositions of Cu-SSZ-13 samples synthetized and characterized by FTIR and acid sites titration by Paolucci et al. in the original work.75 The two compositions selected as representative of Cu-species at 1Al and 2Al sites and further characterized by XAS, part (c), are highlighted by white circles. (b) FTIR spectra of Cu-SSZ-13 samples with Si/Al = 15 and variable Cu/Al ratio in the 0 − 0.44 range, collected at 200 °C after of O2-activation at 400 °C. The dashed arrows indicate the increase of the fingerprint band assigned to 1Al Z[Cu(II)OH] complexes and the simultaneous decrease of the Brønsted acid sites bands, as Cu/Al ratio increases. (c) Cu K-edge XANES spectra collected on the two samples representative of Cu-species at 1Al and 2Al sites (top and bottom panels, respectively) after treatment in 20% O2 at 400 °C (solid blue lines), He at 400 °C (dashed teal lines), and in 3% H2 at (dot-dash red lines). Adapted by permission of the American Chemical Society (copyright 2016) from ref.75
Firstly, the authors employed DFT to rank the stability of different Cu-species in the CHA framework. They found the free energy associated with Cu at 2Al sites to be significantly lower than that for Cu near 1Al at both 25 °C and 400 °C. In line with TPR findings described in Section 2.2.1 (Fig. 10), 2Al exchange sites in 6r are thus predicted to represent preferential locations for Cu(II) ions, over a wide interval of conditions. The Al distribution for a given Si/Al ratio in the framework was then determined by numerical simulations149 imposing random Al siting subject to the
Löwenstein’s rule.150 Based on the results of computational analysis, all the 2Al sites available at a
fixed Si/Al are assumed to be saturated, before Z[Cu(II)OH] species are formed at 1Al sites. Under these hypotheses, the authors computed the compositional phase diagram for Cu-sites in O2-activated
Cu-SSZ-13, reported in Fig. 11a. It shows the fraction of Cu occurring as 1Al Z[Cu(II)OH] as a function of Si/Al and Cu/Al ratios. Below the white line, 2Al Z2Cu(II) is predicted to be the only
Cu-species present, while above the line the fraction of Z[Cu(II)OH] progressively increases, becoming largely dominant in the right top corner of the compositional plane. Bates et al.151 previously
ratio under similar assumptions. They predicted 2Al saturation at Cu/Al ≈ 0.24, 0.09, and 0.05 for Si/Al = 5, 15 and 29, respectively.
Schneider and co-workers reported several experimental results to validate the theoretical diagram in Fig. 11a. They synthesized a large series of Cu-SSZ-13 catalysts with Si/Al = 5, 15, 25 and various Cu-loadings (full white dots in Fig. 11a). The samples were characterized by Brønsted acid sites titration as well as in situ FTIR (Fig. 11b) and XAS spectroscopy (Fig. 11c) for selected catalysts. Cu/H+ exchange stoichiometries obtained quantifying the residual H+ in the exchanged materials by
NH3-TPD, revealed that Cu ion exchange occurs in a sequential way: first as Z2Cu(II) [Z2H2 +
Cu(II)→ Z2Cu(II) + 2H+] up to saturation of available 2Al sites and then as Z[Cu(II)OH] at 1Al sites
[ZH + Cu(II)H2O→ Z[Cu(II)OH] + 2H+]. Further spectroscopic evidence on this point came from
the FTIR data shown in Fig. 11b. Here, the Z[Cu(II)OH] fingerprint band at ca. 3650 cm1 (see also
Section 2.1) remains undetected until a threshold in Cu-loading is reached, corresponding to Cu/Al = 0.21 for the investigated samples with Si/Al = 15. For Cu/Al > 0.21, a progressive increase in the 3650 cm1 band intensity is observed. Complementary insights were obtained by monitoring the
characteristic vibrations of Brønsted sites at 3605 and 3580 cm−1.63, 152 In particular, by comparing
the integrated peak areas for these two bands in the Cu-exchanged and protonic zeolites, a 2:1 H+/Cu
ratio was found until Cu/Al = 0.12, while a 1:1 H+/Cu was determined for Cu/Al ≥ 0.21.
In situ XAS finally allowed the authors to directly probe the redox behaviour and local coordination environment of Cu for two selected compositional points (white circles in Fig. 11a), representative of 1Al and 2Al Cu-species. Fig. 11c shows the Cu K-edge XANES spectra for the two catalysts collected after high-temperature treatment under oxidant (20% O2/He) and reducing
atmosphere (both He and diluted H2). While treatment in oxygen results in a largely dominant Cu(II)
oxidation state in both the samples, their response to reducing conditions is drastically different. The 2Al sample only undergoes very minor modifications, preserving the characteristic Cu(II) XANES features even in H2. Conversely, the 1Al sample undergoes substantial reduction after treatment in
both He (self-reduction process) and H2. Corresponding EXAFS spectra collected for 1Al and 2Al
samples in O2 and He were in qualitative agreement with the model structures of Z[Cu(II)OH]/ZCu(I)
(O2/He) and Z2Cu(II) (both O2 and He), respectively.
2.2.3. Further spectroscopic validation, quantification and deviations from the ideal picture. Overall, the results reviewed in the previous section represent a qualitative fundamental contribution to our current understanding of composition-dependent Cu-speciation in Cu-SSZ-13. Redox-active Z[Cu(II)OH] and redox-inert Z2Cu(II) species emerge as the two key players in the field, with the
latter preferentially stabilized at low values of both Si/Al and Cu/Al ratios, due to statistical availability of suitable 2Al docking sites and more favourable energetics. However, open questions remained on how these framework-interacting Cu-species are formed from the fully hydrated Cu(II) aquo complexes known to dominate in the as-prepared materials (see Section 2.1). Even more important, the need for a more quantitative evaluation of the experimental results was highly desirable since this was often hampered by the co-existence of several rather similar species.
Aiming at a comprehensive experimental exploration of the composition effects on Cu-speciation and (self-) reducibility in Cu-CHA, Martini et al.59 monitored by in situ XANES the He-activation
process from room temperature (RT) to 400 °C on a series of six Cu-SSZ-13 samples with Si/Al ratios in the 5–29 range and Cu/Al ratios from 0.1 to 0.6. Thermal treatment in inert atmosphere was preferred in order to achieve a better spectroscopic contrast between redox-active and redox-resistant Cu-sites, as well as to gain deeper insights in the self-reduction process. High-quality EXAFS spectra on the whole sample series were also collected upon stabilization at 400 °C in He.
Principal component analysis (PCA) of the in situ XANES dataset in Fig. 12a revealed the presence of five principal components (PCs). Thus, the authors applied a multivariate curve resolution (MCR) procedure, based on alternating least square (ALS) method153-155 to extract chemically
meaningful spectra (Fig. 12b) and concentration profiles (Fig. 12c) of the five ‘pure’ Cu-species highlighted by PCA, as a function of temperature and compositional parameters.
The theoretical XANES spectra obtained from MCR-ALS are in excellent agreement with previous XAS studies on Cu-SSZ-13.53, 58, 75, 156. Based on the spectroscopic fingerprints of each
theoretical XANES component and the corresponding temperature-dependent concentration profiles, it was possible to reliably assign each pure spectrum to the Cu-species shown in Fig. 12d. The assignment was further corroborated by XANES simulations computed from the DFT-optimized geometries of the proposed Cu-species. Notably, the MCR spectra attributed to Z[Cu(II)OH] and Z2Cu(II) (black and orange curves in Fig. 12b, respectively) are in perfect agreement with the XANES
reported by Paolucci et al.75 for O
2-treated catalysts representative of Cu at 1Al and 2Al sites, shown
here in Fig. 11c.
Fig. 12. (a) Experimental temperature-dependent in situ XANES spectra collected on six Cu-SSZ-13 samples with different composition (denoted by (Cu/Al; Si/Al) labels) during thermal treatment from 25 to 400 °C (heating rate: 5 °C/min) in 100 ml/min of pure He. Thick dashed lines: starting spectrum at RT; grey thin lines: intermediate states during thermal treatment, thick solid lines: final state at 400 °C. The global dataset includes 72 spectra. (b) Theoretical XANES spectra of the five pure components derived from MCR-ALS analysis. The inset reports a magnification of the pre-edge region in the theoretical spectra. (c) Bar plots reporting temperature-dependent concentration profiles of pure species for the six investigated compositional points. (d) Proposed assignment of the five pure components to specific Cu-species/sites formed in the catalyst as a function of composition and activation temperature. Atom colour code in the structures: Cu, green; H, white; O, red, Si, grey; Al, yellow. In parts b-d, the same colour code is used to indicate the identified pure Cu-species. Unpublished Figure, reporting data previously published in ref.59.
These results demonstrated the potential of XAS spectroscopy combined with multivariate data modelling in tackling the structural complexity associated with Cu-speciation in Cu-CHA. Moreover, the study provided novel insights into the temperature/time-dependent dynamics yielding to framework interacting Cu-sites in the cages of the CHA zeolite during the dehydration process. In particular, the formation of framework-interacting Cu-species from the mobile Cu(II) aquo-complexes present at RT is observed to occur via a four-coordinated Cu(II) dehydration intermediate, peaking around 130 °C (green curve and bars in Fig. 12b,c). Then, Z[Cu(II)OH] and Z2Cu(II) species
progressively develop, with relative abundance determined by composition. Z[Cu(II)OH] peaks appear around 200 °C and thereafter progressively decrease, in favour of self-reduced ZCu(I) species. Conversely, Z2Cu(II) sites, dominant at Si/Al = 5 and favored by low Cu/Al values, reach a steady
population in the 200300 °C range and remain stable until 400 °C. Cu-speciation at 400 °C can be described for all samples as a combination of redox-active Cu-species at 1Al sites (in their oxidized, Z[Cu(II)OH], or reduced, ZCu(I), form) and redox-inert Z2Cu(II) species at 2Al sites in 6r, as
independently validated by DFT-assisted multi-component EXAFS fits in the original work.59
The quantitative knowledge of Cu-speciation enabled from MCR analysis of in situ XANES provided further spectroscopic support to the compositional phase diagram by Paolucci et al.75
However, it also highlighted intriguing deviations from the ideal picture, herein examined in critical comparison with other recent results appeared in the literature.
2.2.3.1. Composition impact on self-reducibility and nature of ZCu(I) species. A first consideration concerns the redox behavior observed by Martini et al.59 for Cu-SSZ-13 with Si/Al = 19 and 29. Not
surprisingly, at such high Si/Al ratios, redox-resistant Z2Cu(II) species represent a minor contribution
to speciation, becoming barely detectable at Si/Al = 29 (Fig. 12c). Although in these catalysts Cu-speciation is dominated by Z[Cu(II)OH] species, it appears that, at such high Si/Al values, self-reduction to ZCu(I) is hampered. These evidences suggest that proximity and spatial distribution of Z[Cu(II)OH] complexes and Brønsted sites within the CHA framework could represent a critical factor in the reduction mechanism. The global picture about the composition impact on self-reducibility in Cu-SSZ-13 for the key compositional points investigated by Martini et al.59 is
summarized in Fig. 13a. The bar plot reports the fractions of Cu-species evaluated from MCR-ALS analysis of the XANES spectra collected 400 °C in He, at the end of the thermal treatment.
At low Si/Al, redox-resistant Z2Cu(II) species dominates the speciation, while at high Si/Al
self-reduction of Z[Cu(II)OH] only occurs to a limited extent. Consequently, the catalyst self-reducibility, quantified by the measured fraction of ZCu(I), reaches an optimum at intermediate Si/Al 15, and it is overall promoted by high Cu/Al ratios.
Fig. 13. (a) Bar plot summarizing Cu-speciation evaluated from MCR-ALS analysis of the in situ XANES spectra collected 400 °C in He for selected Cu-SSZ-13 samples with different composition, indicated by (Cu/Al; Si/Al) labels. (b) Low temperature ( 160 °C) normalized IR spectra of N2 dosed ad increasing equilibrium pressure (from 102 to 5 Torr; 1 Torr = 133.3 Pa) on the same set of samples considered in part (a), after thermal treatment in vacuum at 400 °C. Black, red and light grey curves refer to lowest, highest and intermediate N2 coverage, respectively. Vertical lines indicate the two IR bands assigned to Cu(I)/N2 adducts: low-frequency (LF) band at 2292 cm1 in light grey; high-frequency (HF) band at 2300 cm1 in grey. (c) Bar plots reporting the normalized areas of the N2/Cu(I) IR bands measured at the highest coverage. (d) DFT-optimized models for 1Al ZCu(I) in 8r (top) and 6r (bottom) connected with the LF and HF components detected in the IR spectra in part (b); atom color code: Cu, green; H, white; O, red; Si, grey; Al, yellow). Unpublished Figure, reporting results previously published in ref.59.
Further insights into the abundance and the nature of ZCu(I) species were obtained by in situ IR spectroscopy of adsorbed N2 on the same series of Cu-SSZ-13 samples. The N2 probe molecule
selectively forms Cu(I)/N2 adducts, stable at liquid nitrogen temperature, –160 °C.45, 65 Although
several precautions should be taken in the comparison with XAS results (different experimental conditions, i.e. thermal treatment in He gas flow vs vacuum and data collection at different
